How to set window size of DNA

Peter Sibbald sibbald at
Thu Jun 15 21:12:51 EST 1995

re: "right" window size for seeking DNA correlations

here is one opinion..

You can adopt one of two stances. You can decide what question you are
trying to answer and then set the window size to something appropriate.
For example, you might be asking if the wrapping of DNA on nucleosomes
caused repeated patterns and might choose a window of 160 (if memory ser
ves). OR you can decide that you have no idea what question you are
asking; you merely wonder if there are any patterns in the data to
be discovered. This smacks of exploratory data analysis (EDA). In
this case you should search as much space as you can until you find
a pattern and then try to understand/explain it. 

As references, see papers by Trifinov (in Israel) who has done some
nice work on this. I was also co-author on a couple of papers published
a while back in J. theor Biol. which looked at window sizes among
other things.

Peter R. Sibbald 

More information about the Comp-bio mailing list